t/t{3600,3800,5401}: do not use egrep when grep would do
[git] / builtin-fast-export.c
1 /*
2  * "git fast-export" builtin command
3  *
4  * Copyright (C) 2007 Johannes E. Schindelin
5  */
6 #include "builtin.h"
7 #include "cache.h"
8 #include "commit.h"
9 #include "object.h"
10 #include "tag.h"
11 #include "diff.h"
12 #include "diffcore.h"
13 #include "log-tree.h"
14 #include "revision.h"
15 #include "decorate.h"
16 #include "path-list.h"
17 #include "utf8.h"
18 #include "parse-options.h"
19
20 static const char *fast_export_usage[] = {
21         "git-fast-export [rev-list-opts]",
22         NULL
23 };
24
25 static int progress;
26 static enum { VERBATIM, WARN, STRIP, ABORT } signed_tag_mode = ABORT;
27
28 static int parse_opt_signed_tag_mode(const struct option *opt,
29                                      const char *arg, int unset)
30 {
31         if (unset || !strcmp(arg, "abort"))
32                 signed_tag_mode = ABORT;
33         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
34                 signed_tag_mode = VERBATIM;
35         else if (!strcmp(arg, "warn"))
36                 signed_tag_mode = WARN;
37         else if (!strcmp(arg, "strip"))
38                 signed_tag_mode = STRIP;
39         else
40                 return error("Unknown signed-tag mode: %s", arg);
41         return 0;
42 }
43
44 static struct decoration idnums;
45 static uint32_t last_idnum;
46
47 static int has_unshown_parent(struct commit *commit)
48 {
49         struct commit_list *parent;
50
51         for (parent = commit->parents; parent; parent = parent->next)
52                 if (!(parent->item->object.flags & SHOWN) &&
53                     !(parent->item->object.flags & UNINTERESTING))
54                         return 1;
55         return 0;
56 }
57
58 /* Since intptr_t is C99, we do not use it here */
59 static void mark_object(struct object *object)
60 {
61         last_idnum++;
62         add_decoration(&idnums, object, ((uint32_t *)NULL) + last_idnum);
63 }
64
65 static int get_object_mark(struct object *object)
66 {
67         void *decoration = lookup_decoration(&idnums, object);
68         if (!decoration)
69                 return 0;
70         return (uint32_t *)decoration - (uint32_t *)NULL;
71 }
72
73 static void show_progress(void)
74 {
75         static int counter = 0;
76         if (!progress)
77                 return;
78         if ((++counter % progress) == 0)
79                 printf("progress %d objects\n", counter);
80 }
81
82 static void handle_object(const unsigned char *sha1)
83 {
84         unsigned long size;
85         enum object_type type;
86         char *buf;
87         struct object *object;
88
89         if (is_null_sha1(sha1))
90                 return;
91
92         object = parse_object(sha1);
93         if (!object)
94                 die ("Could not read blob %s", sha1_to_hex(sha1));
95
96         if (object->flags & SHOWN)
97                 return;
98
99         buf = read_sha1_file(sha1, &type, &size);
100         if (!buf)
101                 die ("Could not read blob %s", sha1_to_hex(sha1));
102
103         mark_object(object);
104
105         printf("blob\nmark :%d\ndata %lu\n", last_idnum, size);
106         if (size && fwrite(buf, size, 1, stdout) != 1)
107                 die ("Could not write blob %s", sha1_to_hex(sha1));
108         printf("\n");
109
110         show_progress();
111
112         object->flags |= SHOWN;
113         free(buf);
114 }
115
116 static void show_filemodify(struct diff_queue_struct *q,
117                             struct diff_options *options, void *data)
118 {
119         int i;
120         for (i = 0; i < q->nr; i++) {
121                 struct diff_filespec *spec = q->queue[i]->two;
122                 if (is_null_sha1(spec->sha1))
123                         printf("D %s\n", spec->path);
124                 else {
125                         struct object *object = lookup_object(spec->sha1);
126                         printf("M 0%06o :%d %s\n", spec->mode,
127                                get_object_mark(object), spec->path);
128                 }
129         }
130 }
131
132 static const char *find_encoding(const char *begin, const char *end)
133 {
134         const char *needle = "\nencoding ";
135         char *bol, *eol;
136
137         bol = memmem(begin, end ? end - begin : strlen(begin),
138                      needle, strlen(needle));
139         if (!bol)
140                 return git_commit_encoding;
141         bol += strlen(needle);
142         eol = strchrnul(bol, '\n');
143         *eol = '\0';
144         return bol;
145 }
146
147 static void handle_commit(struct commit *commit, struct rev_info *rev)
148 {
149         int saved_output_format = rev->diffopt.output_format;
150         const char *author, *author_end, *committer, *committer_end;
151         const char *encoding, *message;
152         char *reencoded = NULL;
153         struct commit_list *p;
154         int i;
155
156         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
157
158         parse_commit(commit);
159         author = strstr(commit->buffer, "\nauthor ");
160         if (!author)
161                 die ("Could not find author in commit %s",
162                      sha1_to_hex(commit->object.sha1));
163         author++;
164         author_end = strchrnul(author, '\n');
165         committer = strstr(author_end, "\ncommitter ");
166         if (!committer)
167                 die ("Could not find committer in commit %s",
168                      sha1_to_hex(commit->object.sha1));
169         committer++;
170         committer_end = strchrnul(committer, '\n');
171         message = strstr(committer_end, "\n\n");
172         encoding = find_encoding(committer_end, message);
173         if (message)
174                 message += 2;
175
176         if (commit->parents) {
177                 parse_commit(commit->parents->item);
178                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
179                                commit->tree->object.sha1, "", &rev->diffopt);
180         }
181         else
182                 diff_root_tree_sha1(commit->tree->object.sha1,
183                                     "", &rev->diffopt);
184
185         for (i = 0; i < diff_queued_diff.nr; i++)
186                 handle_object(diff_queued_diff.queue[i]->two->sha1);
187
188         mark_object(&commit->object);
189         if (!is_encoding_utf8(encoding))
190                 reencoded = reencode_string(message, "UTF-8", encoding);
191         printf("commit %s\nmark :%d\n%.*s\n%.*s\ndata %u\n%s",
192                (const char *)commit->util, last_idnum,
193                (int)(author_end - author), author,
194                (int)(committer_end - committer), committer,
195                (unsigned)(reencoded
196                           ? strlen(reencoded) : message
197                           ? strlen(message) : 0),
198                reencoded ? reencoded : message ? message : "");
199         if (reencoded)
200                 free(reencoded);
201
202         for (i = 0, p = commit->parents; p; p = p->next) {
203                 int mark = get_object_mark(&p->item->object);
204                 if (!mark)
205                         continue;
206                 if (i == 0)
207                         printf("from :%d\n", mark);
208                 else if (i == 1)
209                         printf("merge :%d", mark);
210                 else
211                         printf(" :%d", mark);
212                 i++;
213         }
214         if (i > 1)
215                 printf("\n");
216
217         log_tree_diff_flush(rev);
218         rev->diffopt.output_format = saved_output_format;
219
220         printf("\n");
221
222         show_progress();
223 }
224
225 static void handle_tail(struct object_array *commits, struct rev_info *revs)
226 {
227         struct commit *commit;
228         while (commits->nr) {
229                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
230                 if (has_unshown_parent(commit))
231                         return;
232                 handle_commit(commit, revs);
233                 commits->nr--;
234         }
235 }
236
237 static void handle_tag(const char *name, struct tag *tag)
238 {
239         unsigned long size;
240         enum object_type type;
241         char *buf;
242         const char *tagger, *tagger_end, *message;
243         size_t message_size = 0;
244
245         buf = read_sha1_file(tag->object.sha1, &type, &size);
246         if (!buf)
247                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
248         message = memmem(buf, size, "\n\n", 2);
249         if (message) {
250                 message += 2;
251                 message_size = strlen(message);
252         }
253         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
254         if (!tagger)
255                 die ("No tagger for tag %s", sha1_to_hex(tag->object.sha1));
256         tagger++;
257         tagger_end = strchrnul(tagger, '\n');
258
259         /* handle signed tags */
260         if (message) {
261                 const char *signature = strstr(message,
262                                                "\n-----BEGIN PGP SIGNATURE-----\n");
263                 if (signature)
264                         switch(signed_tag_mode) {
265                         case ABORT:
266                                 die ("Encountered signed tag %s; use "
267                                      "--signed-tag=<mode> to handle it.",
268                                      sha1_to_hex(tag->object.sha1));
269                         case WARN:
270                                 warning ("Exporting signed tag %s",
271                                          sha1_to_hex(tag->object.sha1));
272                                 /* fallthru */
273                         case VERBATIM:
274                                 break;
275                         case STRIP:
276                                 message_size = signature + 1 - message;
277                                 break;
278                         }
279         }
280
281         if (!prefixcmp(name, "refs/tags/"))
282                 name += 10;
283         printf("tag %s\nfrom :%d\n%.*s\ndata %d\n%.*s\n",
284                name, get_object_mark(tag->tagged),
285                (int)(tagger_end - tagger), tagger,
286                (int)message_size, (int)message_size, message ? message : "");
287 }
288
289 static void get_tags_and_duplicates(struct object_array *pending,
290                                     struct path_list *extra_refs)
291 {
292         struct tag *tag;
293         int i;
294
295         for (i = 0; i < pending->nr; i++) {
296                 struct object_array_entry *e = pending->objects + i;
297                 unsigned char sha1[20];
298                 struct commit *commit = commit;
299                 char *full_name;
300
301                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
302                         continue;
303
304                 switch (e->item->type) {
305                 case OBJ_COMMIT:
306                         commit = (struct commit *)e->item;
307                         break;
308                 case OBJ_TAG:
309                         tag = (struct tag *)e->item;
310                         while (tag && tag->object.type == OBJ_TAG) {
311                                 path_list_insert(full_name, extra_refs)->util = tag;
312                                 tag = (struct tag *)tag->tagged;
313                         }
314                         if (!tag)
315                                 die ("Tag %s points nowhere?", e->name);
316                         switch(tag->object.type) {
317                         case OBJ_COMMIT:
318                                 commit = (struct commit *)tag;
319                                 break;
320                         case OBJ_BLOB:
321                                 handle_object(tag->object.sha1);
322                                 continue;
323                         }
324                         break;
325                 default:
326                         die ("Unexpected object of type %s",
327                              typename(e->item->type));
328                 }
329                 if (commit->util)
330                         /* more than one name for the same object */
331                         path_list_insert(full_name, extra_refs)->util = commit;
332                 else
333                         commit->util = full_name;
334         }
335 }
336
337 static void handle_tags_and_duplicates(struct path_list *extra_refs)
338 {
339         struct commit *commit;
340         int i;
341
342         for (i = extra_refs->nr - 1; i >= 0; i--) {
343                 const char *name = extra_refs->items[i].path;
344                 struct object *object = extra_refs->items[i].util;
345                 switch (object->type) {
346                 case OBJ_TAG:
347                         handle_tag(name, (struct tag *)object);
348                         break;
349                 case OBJ_COMMIT:
350                         /* create refs pointing to already seen commits */
351                         commit = (struct commit *)object;
352                         printf("reset %s\nfrom :%d\n\n", name,
353                                get_object_mark(&commit->object));
354                         show_progress();
355                         break;
356                 }
357         }
358 }
359
360 int cmd_fast_export(int argc, const char **argv, const char *prefix)
361 {
362         struct rev_info revs;
363         struct object_array commits = { 0, 0, NULL };
364         struct path_list extra_refs = { NULL, 0, 0, 0 };
365         struct commit *commit;
366         struct option options[] = {
367                 OPT_INTEGER(0, "progress", &progress,
368                             "show progress after <n> objects"),
369                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
370                              "select handling of signed tags",
371                              parse_opt_signed_tag_mode),
372                 OPT_END()
373         };
374
375         /* we handle encodings */
376         git_config(git_default_config);
377
378         init_revisions(&revs, prefix);
379         argc = setup_revisions(argc, argv, &revs, NULL);
380         argc = parse_options(argc, argv, options, fast_export_usage, 0);
381         if (argc > 1)
382                 usage_with_options (fast_export_usage, options);
383
384         get_tags_and_duplicates(&revs.pending, &extra_refs);
385
386         prepare_revision_walk(&revs);
387         revs.diffopt.format_callback = show_filemodify;
388         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
389         while ((commit = get_revision(&revs))) {
390                 if (has_unshown_parent(commit)) {
391                         struct commit_list *parent = commit->parents;
392                         add_object_array(&commit->object, NULL, &commits);
393                         for (; parent; parent = parent->next)
394                                 if (!parent->item->util)
395                                         parent->item->util = commit->util;
396                 }
397                 else {
398                         handle_commit(commit, &revs);
399                         handle_tail(&commits, &revs);
400                 }
401         }
402
403         handle_tags_and_duplicates(&extra_refs);
404
405         return 0;
406 }